4 ms·
I'm curious how this will perform vs AMD's Epyc line in terms of performance per Watt on different workloads.
by tmikaeld 7y ago
I'm curious how this will perform vs AMD's Epyc line in terms of performance per Watt on different workloads.
- drewg123 7y agoIt depends on the workload. We tried the ampere emag, and what killed it for us was that TLS performance was nowhere near modern x86-64 CPUs (Intel or AMD)
- cameron_b 7y agodid you use them on Packet?
- magicalhippo 7y agoAs a n00b, is that down to poor or lacking encryption hardware support ala AES-NI? Or something else?
- drewg123 7y agoYes.
- floatboth 7y agoYou did make sure the AES instructions were used, right? I wouldn't be surprised the AES unit on the eMAG is relatively slow — other units seem to be as well, e.g. in my silly CRC32 benchmark, the Arm Cortex-A72 did 1kb in 79 ns at just 2.0 GHz while the eMAG did it in 103 ns at much faster clocks (3.0 or 3.3 GHz). I suspect they might have reused these HW blocks from the old Applied Micro X-Gene :D But now on the new product, it's all Arm Neoverse cores, it's gonna be great.
- Rebelgecko 7y agoWas that heavily cipher dependent? I wouldn't be surprised if Chacha20 performed much better than AES w/o any hardware acceleration (other than SIMD instructions)
- drewg123 7y agoAh, that's interesting. We don't use chacha20. This was AES-GCM
- Rebelgecko 7y agoThe situation is probably better now that ARMv8 has some crypto-specific instructions, but AES-GCM on older ARMs performed awfully without the instructions specifically for doing AES and Galois field multiplication